“This is the way the world ends. Not with a bang but a whimper.”
T. S. Eliot
Today I want to talk about the existential threat from artificial intelligence, not in the apocalyptic sense, but in the philosophical existentialist sense. Not too long ago, Large Language Models like ChatGPT were not able to tell which of 9.11 or 9.9 is the larger number. Now, they can solve PhD qualifying exam level problem in seconds. Benchmarks for state of the art AIs like Humanity’s Last Exam and Frontier Math now include mathematics problems that are hard even for experts. Following the current trajectory, reasoning models in the future will most likely synergize with formal proof assistants like Lean and Coq, and will probably train on an endless supply of synthetically generated mathematics – there seems to be unlimited potential. It is surreal to think that all of these progress has unfolded so recently. To be perfectly honest, the meteoric rise of AI has struck me with both awe and fear, but mostly the latter. As mathematicians, we suddenly find ourselves confronting the grim possibility that AI might one day reach the frontiers of research or even beyond – it is like the sword of Damocles, a spectre looming in the background. My fear is not the cliched scifi trope of some Skynet-like AI obliterating humanity. Rather, my fear is that long before AI poses any physical threat — if it ever does — it will crush our senses of meaning and purpose, that they will destroy us spiritually way before they do physically.
This experience is not unique to mathematicians. After the advant of diffusion models, I see many artists expressing their existential frustration about AI art, and the same happened with musicians and novelists. Years after AlphaGo defeated the best human Go player, top Go player Ke Jie also expressed his existential confusion in a talk. It should be noted that all these anxieties are not simply that AI might “take our jobs”. No, it is something much deeper. To many of these people, what they do constitutes a large part of their meaning for existence. They pride themselves on this attitude and they appreciate this in each other. The confrontation with AI has forced them into an uncomfortable philosophical quandary: either they concede that AIs shares their existential stature as creative beings (and often better ones at that), or they accept the nihilistic and painful indictment that their craft holds little inherent meaning. This is the ultimate melancholy. How do we grapple with this dilemma? Will we ever be able to find a modus vivendi with AI? This question has been haunting me for the last few weeks.
Existentialism is a reaction to nihilism caused by the crisis of modernity. Following the Enlightenment, modern science irrevocable devastated the premodern existential narrative based on religious faith. This is epitomized by Nietzsche’s famous proclaimation, “God is dead”. As a result, for the industralized world, more and more people have been placing their purposes and meanings of life to personal, often creative, endeavours, instead of a devotion to religion. One of the most important ideas of Jean-Paul Sarte, one of the greatest French existentialist philosophers of the last century, is l’existence précède l’essence, i.e. existence precedes essence. The essence of a thing, in Sartrean jargon, is the attributes of which that characterizes its existence – the essence of an object is what makes it what it is. A chair, for example, is characterized by the attributes of “chairness” (such as to provide a surface on which to sit) as its predetermined essence or purpose. By saying that existence precedes essence, Sarte argues that we do not exist due to any predetermined inherent essence – we have to create our own purposes. This could mean many activities depending on the person (art, music, math being some examples).
However, I believe that the emergence of AI challenges precisely this narrative. We find that AI is capable of performing the very activities we find meaningful for ourselves, all the while being staunchly inhuman – it remains something created for the purpose of serving humans, an object with an essence. For me, this gives me a feeling of horror similar to the uncanny valley. It is the feeling of something inhuman is somehow more human than me. Hence the aforementioned dilemma: either AI is truly as human as I am, or these activities are not really all that meaningful. Realistically speaking, I don’t seriously think AI could completely replace human mathematicians, so there should still be meaning in doing mathematics in the future, albeit probably with the help of AI as some super advanced logico-calculator. However, for people who are affected by AI in this way, myself included, I think AI compels them to rethink profoundly what makes their lives meaningful.
Comments